Nov 11AI vendor breach exposes Notion AI chat transcripts.Why it matters

AI at the core of how you run.

We rebuild the workflows where people read, sort, check, and decide, with AI doing the heavy lifting and your team staying in control. You choose who builds it and who runs it.

Built and run by DataAutomation

YouCan we approve the Acme Supply quote that came in by email?

Emailquote PDF read: $4,820 for 600 unitsread
Purchasinglast PO price $7.40 a uniton file
Accountingvendor terms Net 30ok
DA ruleprice 9% over last PO, needs approvalheld
DataAutomation assistantNot yet. The quote works out to $8.03 a unit, about 9% above the last PO, so I held it for your approval. Approve at the new price, or ask Acme to match?
<15 days

to the first AI win

We start with one workflow that pays back quickly, prove it on your real data, then expand.

Most AI projects stall between the demo and the real business. The prompt that worked on three clean examples meets messy emails, PDFs, and edge cases, and nobody owns it after launch.

We put AI into the workflows where people read, sort, check, and decide, on top of systems we know how to connect. Rules handle the predictable parts, AI handles the judgment calls, and people approve what you want approved.

Where it earns its keep

AI earns its keep where inputs are messy and volume is real: wholesale orders by email, supplier documents, customer messages, and data that needs checking before it moves.

It also earns its keep in leadership: weekly briefings that build themselves, margin analysis, and decisions made on your own numbers instead of opinions.

Common use cases

What clients usually start with.

Emailed orders read automatically

Orders in emails and PDFs turned into clean orders in your system, with a person approving anything unclear.

Hours of data entry back every week.

Supplier documents checked

Confirmations, price lists, and invoices compared to your POs and catalog.

Mistakes caught before they cost money.

Customer messages drafted

Replies drafted from your real policies and order data.

Faster answers that stay accurate.

A business briefing that writes itself

Sales, margin, cash, and operations summarized every week.

Leaders start with decisions.

Data checked before it moves

AI flags records that look wrong before they reach accounting or production.

Fewer cleanups later.

Your team trained to build

An internal champion who can build and extend AI workflows.

Capability that stays in the business.

Where it gets complicated

What usually goes wrong, and what we do about it from day one.

  • Demos that never meet real data

    A prompt that works on three clean examples is not a system. Real emails, PDFs, and orders are messy, and that is where pilots stall.

    We build and test on your messiest real inputs before anything goes live.

  • AI where a rule would do

    If something can be written down as a rule, a rule is cheaper, faster, and never has an off day. AI everywhere is expensive and harder to trust.

    We use code for the predictable parts and AI only for the judgment calls.

  • Nobody owns it after launch

    AI workflows need someone watching the exceptions, especially in the first month. Without an owner they drift and get switched off.

    We agree who owns it before we build, and give them a simple queue for anything unclear.

  • Customer data pasted into chats

    Teams paste orders, customer lists, and passwords into AI tools nobody approved. In IBM’s 2026 report, that kind of shadow AI was involved in 43% of breaches.

    We keep AI inside approved, audited connections that pass only the fields a task needs.

What you get

  • A ranked list of AI opportunities

    Where your team spends time reading, retyping, and checking, with the payback of each.

  • Your first workflow, live

    Built on your data, with a clear queue for anything that needs a person.

  • Measured results

    Time saved and errors caught, tracked against the goal we set at kickoff.

  • Guardrails where you need them

    Access limits, logging, and approvals set to how much you want AI to do on its own.

  • Your team, trained

    People who can run it, adjust it, and spot where AI should go next.

  • A roadmap

    The next workflows in order, with what each will take.

What a typical project looks like

  1. Weeks 1 to 2: find the first win

    We map where time goes and pick the workflow with the fastest payback. No separate scoping fee.

  2. Weeks 2 to 4: build on your data

    Rules for the routine cases, AI for the messy ones, and a review queue for your team.

  3. First month: prove it

    We measure against the agreed goal. On Teach and Run, if we miss it, you get that month back.

  4. After that: expand

    The next workflow, and the one after, at the pace your plan sets.

Our full process →

Is this right for you?

A good fit if

  • People on your team spend hours reading, sorting, or retyping
  • You have tried AI tools and nothing stuck in production
  • You want AI inside the systems you already run, not another app

Probably not the right fit if

  • The work is fully predictable and rules-based (an integration costs less)
  • You want an off-the-shelf tool, not something built around your workflow
  • You need it live this week without testing on real data

Pricing

Teach

You build. We teach. Your team builds with our engineers alongside, and owns it at the end. Includes our first-month guarantee.

$7,000 / month

Guide

We build. You decide. A guided 3-month program: weekly working sessions with your leadership, an AI roadmap, and at least one workflow live by the end of the term, handed over with training and documentation.

$6,900 / 3-month term

Run

Managed AI. We build. We run. We build it and run it in your environment, with production support for everything we ship. Entry $5,000 a month: at least 1 workflow shipped a month. Growth $10,000: at least 3 a month. Continuous $15,000: at least 1 a week. Includes our first-month guarantee.

From $5,000 / month

See all pricing →

A note on trust

We have been audited by Deloitte and PwC, and we complete Amazon's SP-API security assessment. Your data stays in your systems, approvals are set to how much you want automated, and every automated step is logged. See our audits →

Questions

Do you only do AI?

No. Most of our work is code, and AI is used where it actually helps. Our integrations often include an AI step for the messy cases.

Can we opt out of AI?

Yes. If you would rather not use AI, tell us and we build it with code and people instead.

How fast do we see results?

Our first AI win typically lands in under 15 days.

What does it start with?

No separate scoping. The first two weeks of every plan find the workflow with the fastest payback, then we build it.

Is there a guarantee?

Yes, on Teach and Run. We agree on a measurable goal at kickoff. If we do not hit it in the first month, you get that month back.

Which AI models do you use?

The ones that fit the job, including Claude and ChatGPT models. We design so you can change models later.

Will our data be used to train AI?

We use business-grade AI services that do not train on your data, and we keep customer personal data out of AI steps unless it is required and allowed.

Can AI place orders, move money, or email customers for us?

Yes. We set the guardrails you want, from a person approving every step to fully hands-off, and we own mistakes in what we run.

Do we need our own AI accounts?

Usually it runs in your environment on business accounts you own, so you keep control of the data and the bill.

Audited every year.

Deloitte and Amazon SP-API audits in 2026. PwC and Amazon SP-API in 2025. Shadow AI was involved in 43% of the breaches IBM studied for its 2026 report; we keep your data out of it. And if something we run goes wrong, we own the fix.

See the latest AI incidentsSee our audits